"Bad response from server: 500" on restart during Capistrano puma:restart

[amencarini]

I'm on puma 2.0.1 behind nginx, with Rails 3.2.12, Ruby 2.0.0, server is Ubuntu Precise

Each time I try to deploy my app via Capistrano, at the deploy:restart hook, or when I try a restart through cap puma:restart I constantly get this:

  * 2013-05-01 11:24:26 executing `puma:restart'
  * executing "cd /var/www/project/current && bundle exec pumactl -S /var/www/project/shared/sockets/puma.state restart"
    servers: ["staging.example.com"]
    [staging.example.com] executing command
 ** [out :: staging.example.com] Bad response from server: 500

The puma server gets restarted correctly but the Capistrano deployment process gets interrupted.

This also happens when I try to stop it:

 * executing "cd /var/www/project/current && bundle exec pumactl -S /var/www/project/shared/sockets/puma.state stop"
    servers: ["staging.example.com"]
    [staging.example.com] executing command
 ** [out :: staging.example.com] Bad response from server: 500

[lephyrius]

Im also seeing this!
Using Ruby 2.0 and Rails 3.2.13 on Debian 6.

[amencarini]

@lephyrius thanks for the feedback, and for reminding me to add Ruby and server info :)

[olistik]

Same error experienced with the same setup described by @amencarini

[netmute]

Same error here. It doesn"t always happen, though. Maybe 50% of the time. The server does not correctly restart after throwing that error, I have to manually start puma again.

[evanphx]

Are you always seeing it from pumactl?

[lephyrius]

For me it is only from restart and stop not stats that gives bad response.
The command: "puma" works.(to start the server)
stats doesn't print out anything and not "bad response" when running.

[amencarini]

@evanphx Capistrano executes this:

bundle exec pumactl -S /var/www/project/shared/sockets/puma.state restart

From what I can see, when I use this to restart, pumactl.sock gets removed, then I get the error, the puma process keeps working and when the whole application becomes available again, pumactl.sock reappears. And yes, I consistently get this, it's not spot. The server is a micro AWS instance: not sure if performance can play a role in this.

[Arie]

I get the same thing. Capistrano 2.15, ruby 1.9.3-p392, puma 2.0.1, rails 3.2.13.

[jockee]

Same here. Capistrano 2.15.4. Rub 2.0.0dev. Puma 2.0.1. Rails 3.2.13.

[macfanatic]

Same problem with the following setup:

capistrano 2.15.4
puma 2.0.1
rails 4.0.0.beta1
ruby 2.0.0p0

[Arie]

I found the backtrace the server shows when this 500 happens for me. Need to do some more investigating to wrap my head around what happens in that file.

Puma caught this error: stream closed
/usr/local/rvm/gems/ruby-1.9.3-p392@foo/gems/puma-2.0.1/lib/puma/server.rb:594:in `write'
/usr/local/rvm/gems/ruby-1.9.3-p392@foo/gems/puma-2.0.1/lib/puma/server.rb:594:in `<<'
/usr/local/rvm/gems/ruby-1.9.3-p392@foo/gems/puma-2.0.1/lib/puma/server.rb:594:in `begin_restart'
/usr/local/rvm/gems/ruby-1.9.3-p392@foo/gems/puma-2.0.1/lib/puma/app/status.rb:34:in `call'
/usr/local/rvm/gems/ruby-1.9.3-p392@foo/gems/puma-2.0.1/lib/puma/server.rb:364:in `handle_request'
/usr/local/rvm/gems/ruby-1.9.3-p392@foo/gems/puma-2.0.1/lib/puma/server.rb:243:in `process_client'
/usr/local/rvm/gems/ruby-1.9.3-p392@foo/gems/puma-2.0.1/lib/puma/server.rb:142:in `block in run'
/usr/local/rvm/gems/ruby-1.9.3-p392@foo/gems/puma-2.0.1/lib/puma/thread_pool.rb:92:in `call'
/usr/local/rvm/gems/ruby-1.9.3-p392@foo/gems/puma-2.0.1/lib/puma/thread_pool.rb:92:in `block in    spawn_thread'
undefined method `split' for nil:NilClass

[amencarini]

I've noticed two different behaviours on two different servers: on the faster server we get the 500 error 50% of the times, on the slower server we get it every time.

[MJLang]

Im getting the same issue - Using Ubuntu 12.10, Ruby 2.0.0, Rails 4.0.0.rc1, Puma at master and Capistrano 2.15.4

[macfanatic]

Has anyone found an easy workaround to this problem yet? For now I have to remove the tmp files for puma & kill the process itself after doing a ps aux | grep puma. I'm not a linux guru, so I might be making that part hard on myself, but I'd still like to use puma with capistrano and this is making it almost impossible.

[rudijs]

I'm also seeing this on Ruby 1.9.3

ruby -v
ruby 1.9.3p392 (2013-02-22 revision 39386) [i686-linux]

rails -v
Rails 3.2.13

/srv/blog/current$ RAILS_ENV=production bundle exec puma -e production -d -b unix:///tmp/blog.sock -S /tmp/blog.state --control 'unix:///tmp/blog_pumactl.sock'

Puma 2.0.1 starting...

• Min threads: 0, max threads: 16
• Environment: production
• Listening on unix:///tmp/blog.sock

/srv/blog/current$ bundle exec pumactl -S /tmp/blog.state restart
Bad response from server: 500

/srv/blog/current$ bundle exec pumactl -S /tmp/blog.state stop
undefined method `split' for nil:NilClass

I'm using rvm so when I try without "bundle exec" I also get same response.

I haven't used it with Capistrano yet so it's likely nothing to do with Capistrano.

Really hoping this bug gets squashed soon I'm turning into a real puma fan.

[zires]

I got same error message!

[tomasmuller]

Same error message and machine setup as @oolyme .

[thetron]

I too am getting this issue, also with the same setup as @tomasmuller and @oolyme. However, I am using capistrano.

[tomasmuller]

Oh, forgot to mention that. I'm also using Capistrano.

[iamroody]

ruby 2.0 rails Rails 3.2.13 , same messages

[doabit]

ruby 2.0.0-p195, padrino and Ubuntu 12.04 , same error,

[davidcelis]

Ruby 2.0.0-p195, Rails 4.0.0.rc1 with Rails-API, and Ubuntu 13.04. Same issue.

[bobbrez]

Same issue on Ruby 2.0.0-p0, Rails 4.0.0.rc1 Ubuntu 13.04 on EC3 Small.

[rebelwarrior]

Trying to deploy a Sinatra app, I believe I'm getting a similar error. Ruby 1.9.3p429 Sinatra 1.4.2 Puma 2.0.1

[amencarini]

@evanphx I'm pretty sure it's unrelated to this, but I've used the master to see if how it went and a new problem arose, that was tackled in pull request #271. I'm using that patch and Capistrano deployment is now alright, at last :) Thanks!

[ryanbooker]

I still get this error with the latest version of puma (2.5.0), which contains both of the patches mentioned here.

puma:restart still can't find the .sock.

I'm using bog standard with no config file by requiring 'puma/capistrano' in my deploy.rb.

puma 2.5.0, capistrano 2.15.5, rails 3.2.14

[waynehoover]

I'm also getting this error with puma 2.5.1, capistrano 2.15.5, rails 4.0.0

I'm not using a config file, just requiring 'puma/capistrano' in deploy.rb

Sometimes it works sometimes it doesn't.

[sydneyitguy]

+1 still getting the same issue with puma 2.5.1, capistrano 2.15.5, rails 4.0.0, ruby 2.0

It works about 20% on linode 1GB plan, 80% 500 error. I can't even stop the server after freezing, so I should use "kill -9" to force kill everything and start puma again.

[tobidelius]

I'm having the same problem as well!